Longitudinal Analyses of Gut Mucosal Microbiotas in Ulcerative Colitis in Relation to Patient Age and Disease Severity and Duration

Abstract

Bacteria belonging to the normal colonic microbiota are associated with the etiology of ulcerative colitis (UC). Although several mucosal species have been implicated in the disease process, the organisms and mechanisms involved are unknown. The aim of this investigation was to characterize mucosal biofilm communities over time and to determine the relationship of these bacteria to patient age and disease severity and duration. Multiple rectal biopsy specimens were taken from 33 patients with active UC over a period of 1 year. Real-time PCR was used to quantify mucosal bacteria in UC patients compared to 18 noninflammatory bowel disease controls, and the relationship between indicators of disease severity and bacterial colonization was evaluated by linear regression analysis. Significant differences were detected in bacterial populations on the UC mucosa and in the control group, which varied over the study period. High clinical activity indices (CAI) and sigmoidoscopy scores (SS) were associated with enterobacteria, desulfovibrios, type E Clostridium perfringens, and Enterococcus faecalis, whereas the reverse was true for Clostridium butyricum, Ruminococcus albus, and Eubacterium rectale. Lactobacillus and bifidobacterium numbers were linked with low CAI. Only E. rectale and Clostridium clostridioforme had a high age dependence. These findings demonstrated that longitudinal variations in mucosal bacterial populations occur in UC and that bacterial community structure is related to disease severity.

INTRODUCTION

Dense bacterial communities colonize the mucus layer covering the surface of the human large bowel (1–6). Most people tolerate this complex antigenic biota, but inflammatory bowel disease (IBD), in the form of ulcerative colitis (UC), develops in about 2 per 10,000 adults every year (7). UC is characterized by intense inflammation of the mucosa, with bloody diarrhea, urgency to defecate, and general malaise (7). While many intestinal pathogens can evoke an acute inflammatory response in the intestinal mucosa (8, 9), the clinical effects of these infections are usually acute rather than chronic, and the pathogenic mechanisms and host responses involved are generally well understood. However, the role of bacteria in other forms of gut disease, such as antibiotic-associated colitis or UC, is less clear.

Although animal studies indicate that bacteria are needed for UC-like symptoms to manifest (10, 11), there does not seem to be a specific transmissible agent involved in the disease process. Moreover, bacteria that have been implicated in UC etiology in various investigations are not found in all patients with the disease (12). Thus, Koch's postulates cannot be demonstrated. Despite this apparent incongruity, bacteria growing on the gut wall may be involved in UC, either as overtly pathogenic organisms colonizing the epithelial surface and invading the underlying mucosa or, alternatively, as nonpathogenic commensal species occupying adhesion sites on the mucosa and preventing the attachment of disease-causing bacteria.

Because mucosal biofilms in the large intestine exist in close association with the host, it is likely that biofilm species interact with the host immune and neuroendocrine system to a greater extent than microorganisms growing in the gut lumen. An increasing body of experimental evidence suggests that bacterial populations colonizing the colonic mucosa are different in composition from those in the gut lumen (3–5, 13–15), but relatively few studies have been made to quantify or characterize mucosal bacteria in UC (3, 4, 14–21).

In this study, we investigated the composition of mucosal bacterial communities in UC patients over a period of 1 year, using real-time PCR, which is a sensitive, quantitative technique that can detect organisms occurring in low numbers on the epithelial surface (16), and related these measurements to clinical information corresponding to UC disease state, as determined by clinical activity indices (CAI), sigmoidoscopy scores (SS), and disease duration.

MATERIALS AND METHODS

Study population.

Thirty-three patients with active UC were recruited for the study from the outpatient clinic of the Gastroenterology Department at Ninewells Hospital, Dundee, United Kingdom. Patients with clinical disease activity and those with disease extending from the proximal to the rectosigmoidal junction, confirmed within the previous 2 years by barium enema, isotope granulocyte scanning, flexisigmoidoscopy, or colonoscopy, were included in the study. Subjects with positive stool cultures of pathogens and antibiotic treatment within 8 weeks were excluded. The patients were treated conventionally and followed up over a period of 1 year. On a 3-monthly visit, rectal biopsy specimens were collected from individuals who had not received any bowel preparation, gently washed, and immediately frozen at −80°C in sterile Eppendorf tubes. A minimum of three biopsy specimens were taken at each time period and analyzed separately. The biopsy specimens were taken from inflamed tissue during the initial visit and from the same area during subsequent visits. Normal rectal biopsy specimens were obtained from patients who were visiting the clinic for altered bowel habit, colorectal screening, previous polyps, and other diagnoses who did not receive antibiotics prior to sample collection. These individuals were determined to be macroscopically normal by the physician who handled the examinations, and pinch biopsy specimens were taken without any bowel preparation. Informed consent was obtained for the investigation, which was approved by the Tayside Committee on Medical Research Ethics, Dundee, United Kingdom.

Clinical activity indices.

Clinical symptoms and sigmoidoscopy scores were used as clinical endpoints. Clinical assessments including history, physical examination, and global clinical grading using Truelove and Witts criteria were done at the time of recruitment and at 3-month intervals during the study. CAI grading was done as described previously by Walmsley et al. (22). This was based on bowel frequency during the day (score of 0 to 3) and at night (score of 1 to 2), urgency of defecation (score of 1 to 3), blood in stools (score of 1 to 3), general well-being (score of 0 to 4), and extracolonic features (arthritis, pyoderma gangrenosum, erythema nodosum, and uveitis, each having a score of 1 point). Sums of the scores were used for data analysis.

Grading of sigmoidoscopic appearance.

Patients were examined by rigid sigmoidoscopy or flexisigmoidoscopy and graded on a scale of 0 to 6, according to the macroscopic appearance of the rectal mucosa at a distance 5 to 10 cm from the anal verge. The sigmoidoscopy grading system was described previously by Baron et al. (23). This included friability (score of 0 or 1), bleeding (score of 0 or 1), vessel pattern (score of 0 to 2), and overall appearance (score of 0 to 2). Sums of the scores were used for data analysis.

Amplification by conventional PCR to check primer specificity.

A Techne Genius PCR machine (Techne Ltd., Duxford, United Kingdom) was used for conventional PCR to check primer specificity. New primers were designed on the basis of toxin gene sequences in the case of Clostridium perfringens, available at the National Center for Biotechnology Information databases and The Ribosomal Database Project. Primer Premier for Windows, version 5.0 (Premier Biosoft International, Palo Alto, CA), was used for searching, aligning, editing, and handling primers for collected bacterial sequences. Primers were purchased from Invitrogen Technologies (Paisley, United Kingdom). PCR consisted of 35 cycles, with an initial DNA denaturation step at 95°C (1 min), followed by annealing at 62°C (1 min) and elongation at 72°C (45 s). The procedure was completed with a final elongation step at 72°C (5 min). Determinations of optimum temperature were done by using a Mastercycler Gradient PCR machine, adjusted for various temperature ranges (Eppendorf-Netherler-Hinz, Hamburg, Germany).

DNA extraction from bacterial cultures.

A range of intestinal isolates and type strains were used as controls for testing specificities of the PCR primers, as described previously (16, 24).

Cell pellets from liquid media or bacterial colonies from plate swabs were resuspended in 450 μl of sterile water and 50 μl of lysozyme (50 mg/ml). Reaction mixtures were incubated at 37°C (30 min). Twenty-five microliters of a proteinase K solution (20 mg/ml), 50 μl 20% SDS, 500 μl of H₂O, and 350 mg glass beads (0.1-mm diameter) were then added, and the mixture was bead beaten before and after incubation in a water bath at 70°C (10 min). Total DNA was obtained after centrifugation (5,000 × g for 3 min).

Bacterial DNA extraction and purification from biopsy material.

Frozen tissue sections were extracted by using QIAmp DNA spin columns (Qiagen Ltd., West Sussex, United Kingdom), as described previously (16). Biopsy specimens (2 to 5 mg) were suspended in lysis buffer and lysozyme solutions before incubation at 55°C. Proteinase K, buffer ATL (Qiagen), and ethanol were added sequentially, and the mixture was bead beaten before and after incubation at 70°C. Bacterial cell lysates were obtained by centrifugation, as described above. Total DNA from bacterial cultures and biopsy specimens was purified by using a DNA purification kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Finally, the purified DNA was eluted in 200 μl of elution buffer provided with the kit.

Real-time PCR.

DNA was amplified by using genus- or species-specific primers to cover a wide range of intestinal bacteria quantified by using an iCycler real-time PCR detection system (Bio-Rad, Hercules, CA) and PCR amplification, and melting curves were analyzed by using iCycler Optical System software, version 3.0 (Bio-Rad). Each reaction was done in duplicate in a volume of 20 μl with 96-well optical-grade PCR plates sealed with optical sealing tape (Bio-Rad). Amplification reactions were done with iQ SYBR green Supermix (Bio-Rad) containing 3 mM MgCl₂, 20 mM Tris HCl (pH 8.4), 50 mM KCl, each deoxynucleotide triphosphate at a concentration of 200 μM, SYBR green 1, 10 nM fluorescein, 0.625 U of iTAQ DNA polymerase mixed with the selected primer set at a concentration of 0.5 μM for each primer, and 1.6 μl of the respective DNA. Amplifications were done with the following temperature profiles: 1 cycle at 95°C for 3 min, 38 cycles of denaturation at 95°C (30 s) followed by 30 s at the appropriate annealing temperature, and one final cycle at 95°C (30 s). The melt curve analysis was run for 43 repeats, from 55°C to 95°C slowly (1°C/cycle of 10 s), as endpoint assays to confirm PCR specificity.

Real-time PCR primers for enumerating mucosal bacteria.

PCR primer sets used for quantitating bacteria in biopsy tissues are shown in Table 1. They were of high specificity and efficiency and were each tested against a range of intestinal bacteria. Repeated measurements of the correlation coefficients and PCR efficiencies for the standard curves at each determination were within the range of 90 to 100%, except for the lactobacillus primer set, which had an E_f of 74%.

Table 1.

Primer sets used for real-time PCR

Target organism	Forward sequence (5′–3′)	Reverse sequence (5′–3′)	Temp (°C)a	Product size (bp)	Reference
Clostridium butyricum	GTGCCGCCGCTAACGCATTAAGTAT	ACCATGCACCACCTGTCTTCCTGCC	72	213	24
Desulfovibrios	CCGTAGATATCTGGAGGAACATCAG	ACATCTAGCATCCATCGTTTACAGC	62	135	16
Clostridium perfringens 1b	ATGTAATAGATAAAGGAGATGGTT	ATAAATTCAGAAGTAAATCCAACT	54	163	25
Clostridium perfringens 2b	CTCATGCTATGATTGTAACTC	GATCATAGGCGTTCTTATCATAATC	55	161	This study
Faecalibacterium prausnitzii	GATGGCCTCGCGTCCGATTAGc	CCGAAGACCTTCTTCCTCC	58	199	26
Clostridium clostridioforme	AATCTTGATTGACTGAGTGGCGGAC	CCATCTCACACTACCGGAGTTTTTC	62	148	24
Peptostreptococcus anaerobius	GCTCGGTGCCTTCACTAACGc	AGCCCCGAAGGGAAGGTGTG	64	188	27
Lactobacilli	GATAGAGGTAGTAACTGGCCTTTAc	GCGGAAACCTCCCAACA	55	390	28
Bifidobacteria	AGGGTTCGATTCTGGCTCAG	CATCCGGCATTACCACCCc	62	156	29
Bacteroides genus	GTCAGTTGTGAAAGTTTGCc	CAATCGGAGTTCTTCGTG	56	127	30
Bacteroides vulgatus	AAGGGAGCGTAGATGGATGTTTA	CGAGCCTCAATGTCAGTTGC	62	192	31
Enterobacteria	CATTGACGTTACCCGCAGAAGAAGC	CTCTACGAGACTCAAGCTTGC	63	195	24
Eubacterium rectale	CGGTACCTGACTAAGAAGC	CCTAGTATTCATCGTTTACGGCGTGb	60	347	32
Ruminococcus albus	CAGGTGTGAAATTTAGGGGC	GTCAGTCCCCCCACACCTAG	63	246	24
Enterococcus faecalis	GCTTTCGGGTGTCGCTGATG	CGTCCTTGTTCTTCTCTAAC	58	253	This study
All eubacteria	ACTCCTACGGGAGGCAGCAGT	GTATTACCGCGGCTGCTGGCAC	54	200	33

^aOptimum melting temperature determined by conventional temperature gradient PCR.

^bPrimer sets for C. perfringens were based on gene sequences other than 16S rRNA: for C. perfringens 1, the cpe gene, encoding iota-toxin component A in C. perfringens type E, was used, and for C. perfringens 2, the alpha-toxin gene in C. perfringens types A to E was used.

^cPrimer modified in this study.

Plasmid DNA standards.

PCR products of the different primer sets were used for preparation of bacterial plasmids for making standard curves, using methods described previously (16). Plasmid DNA was purified by using the Wizard Plus SV Minipreps DNA purification kit (Promega). Plasmid concentrations were determined by electrophoresis and comparison of band strengths against molecular marker DNA (2-log DNA ladder N3200L; New England BioLabs Ltd., Herts, United Kingdom) and by using a spectrophotometer (Cecil Instruments, Cambridge, United Kingdom) at 260/280 nm. DNA concentrations were then converted into 16S rRNA gene copy numbers. Dilutions of the plasmid were used for each real-time PCR assay to generate standard curves for quantitation of target DNA in test samples. DNA melting curves were used to monitor product specificities. Detection was based on fluorescence resonance energy transfer, with a monocolor SYBR green 490 fluorophore. The cycle number at which signal was first detected correlated with the original concentration of the DNA template, and the starting copy number of amplicons was inversely proportional to the real-time threshold cycle. The functional concentration range for the plasmid standards was established to be between molecular copy numbers of 10¹ and 10⁶.

Statistical analysis.

Data analysis was done by using the SPSS v 11.5 Statistics Package (SPSS, Chicago, IL). Copy numbers of 16S rRNA genes per mg of sample were transformed into logarithms, and the normally distributed data were subjected to statistical analysis. One-way analysis of variance (ANOVA) with Bonferroni multiple comparison was used to compare means of bacterial densities between different groups. A P value of <0.05 was considered statistically significant. Linear regression analysis was used to determine the functional relationships of bacterial densities per mg sample against age as well as disease severity and duration. Trend lines of bacterial populations with time were generated by using GraphPad Prism, version 4 (GraphPad Software Inc., San Diego, CA).

Chemicals.

Unless stated otherwise, all chemicals were obtained from Sigma (Poole, Dorset, United Kingdom). Bacteriological culture media were purchased from Oxoid (Basingstoke, Hamps, United Kingdom).

RESULTS

Patients.

Clinical features, including numbers, age, gender, and therapy, of the UC patients involved in this study are shown in Table 2. The mean ages of UC and non-IBD comparators were 53 years and 57.5 years, respectively, UC patients had the disease for a mean period of 14 years, and the majority of these individuals continued their therapy during the study. None of the subjects had received antibiotics in the 12 weeks before commencement of the study.

Table 2.

Characteristics of UC patients^a and non-IBD participants

Characteristic	Value for group
	UC patients	Non-IBD controls
No. of subjects	33	18
Gender (no. of male patients/no. of female patients)	17/16	9/9
Age range (yr) (mean ± SEM)	25–76 (53 ± 0.9)	19–81 (57.5 ± 4.4)
UC duration (yr) (mean ± SEM)	2–40 (13.7 ± 0.83)

^aThe majority of UC patients were receiving the following drugs: mesalamine (Asacol) (14 patients); salazopyrine (5); balsalazide (5); mesalazine (2); combined azathioprine and balsalazide (2); olsalazine (1); and combined prednisolone, azathioprine, and salazopyrine (1). Biopsy specimens were also obtained from six UC patients who were not receiving any type of treatment. Three patients changed therapy from mesalamine to either mesalazine, balsalazide, or salazopyrine during the course of the study. Non-IBD controls received no antibiotics prior to sample collection.

Comparison of mucosal microbiotas in UC patients and non-IBD patients.

The number of copies of group-, genus-, or species-specific 16S rRNA genes or C. perfringens toxin genes determined by real-time PCR provided an indication of the relative sizes of different bacterial populations in mucosal biopsy specimens. Sixteen primer sets for different mucosal bacteria were used to analyze biopsy specimens from the 33 UC patients and 18 patients without IBD (Fig. 1). A significant reduction was found in levels of total eubacteria on the UC mucosa at the start of the study period compared with the non-IBD controls (P < 0.05). Levels of Clostridium clostridioforme, the Eubacterium rectale group, Faecalibacterium prausnitzii, bifidobacteria, lactobacilli, and Clostridium butyricum were also significantly reduced on UC mucosa compared to non-IBD tissues (P < 0.001). Although not as marked, significant reductions were also found for bacteroides, Bacteroides vulgatus, enterobacteria, and desulfovibrios (P < 0.05). Low numbers of C. perfringens 1 and Enterococcus faecalis and increased levels of Peptostreptococcus anaerobius, (P < 0.05) Ruminococcus albus, and C. perfringens 2 were also detected in UC patients compared with the non-IBD controls.

Fig 1.

Comparison of mucosal bacteria in UC patients at baseline (black bars) (n = 33) and in controls (gray bars) (n = 18). Bars represent means of bacterial cell densities, and error bars show standard errors of the means. * denotes a significant difference at a P value of <0.05, and *** denotes a significant difference at a P value of <0.001.

Bacterial populations in UC patients over a 1-year study period.

Changes in levels of the UC microbiotas over 1 year are shown in Fig. 2. Numbers of total eubacteria slightly increased but were generally stable over the study period. Cell population densities of mucosal C. perfringens 1 and 2, Ent. faecalis, P. anaerobius, enterobacteria, and desulfovibrios were relatively stable during the study, while cell densities all of the other bacteria significantly increased (P < 0.01), particularly bifidobacteria, E. rectale, C. clostridioforme, and F. prausnitzii (P < 0.001). A significant increase was observed in levels of B. vulgatus at 6 months compared to baseline levels (P < 0.01), and although marked differences between UC patients and non-IBD controls were observed at the beginning of the study (Fig. 1), this did not hold for the remainder of the investigation. Similarly, levels of F. prausnitzii increased, and while still significantly different from levels in healthy populations at 3 and 9 months (P < 0.05 and P < 0.01, respectively), at 6 and 12 months, they began to reach community sizes detected in the non-IBD controls. Numbers of Clostridium butyricum and C. clostridioforme generally increased during the investigation, although their numbers were always significantly lower than those in non-IBD comparators (P < 0.001). Numbers of lactobacilli increased over the 12 months, although the numbers were still significantly reduced compared to those of the control group at 3 and 6 months (P < 0.001 and P < 0.01, respectively). Levels of enterobacteria (P < 0.001, P < 0.001, and P < 0.01) and the E. rectale group (P < 0.001) were significantly reduced compared to levels in the controls at 3, 6, and 9 months but not at 12 months. Levels of Faecalibacterium prausnitzii showed the greatest increase over the study, increasing from 4.2 log₁₀ ± 0.15 log₁₀ bacterial cells at baseline to 5.8 log₁₀ ± 0.12 log₁₀ bacterial cells at 12 months, and were significantly different from those in control populations only at baseline and 3 months (P < 0.01). Although levels of sulfate-reducing bacteria (SRB) did not change markedly, the significant difference observed compared to levels in non-IBD controls at baseline was not evident after 3 months. Numbers of P. anaerobius were unchanged during the investigation and were significantly different from those in the controls (P < 0.01 at 3 and 6 months, and P < 0.001 at 9 months) for most of the study period. Levels of bacteroides increased over the time period and were significantly lower than in the controls only at baseline.

Fig 2.

Bacterial populations in UC patients over a 1-year time period. Lines are used to show the trends in bacterial populations. Bars represent means of bacterial cell densities, and error bars show standard errors of the means. Patients with no bacteria were excluded from the analysis.

Correlation of bacterial composition with therapy.

The type of anti-inflammatory drug therapy for the UC patients and its effects on total bacterial populations over the time period are shown in Tables 3 and 4. Levels of total eubacteria increased over the 12-month study period in patients taking no medication and in patients on the other UC therapies, with olsalazine, mesalazine, and the combination of azathioprine and balsalazide showing the greatest increases. Several bacterial groups were affected by therapy compared to patients receiving no medication. For instance, the combination of azathioprine and balsalazide was associated with increased levels of enterobacteria (P = 0.05), lactobacilli (P = 0.05), and F. prausnitzii (P = 0.05). Levels of several bacteria, including bifidobacteria (P < 0.05), bacteroides, E. rectale (P < 0.05), C. clostridioforme, and F. prausnitzii (P < 0.05), increased with salazopyrine. The combination of prednisolone, azathioprine, and salazopyrine reduced levels of B. vulgatus (P < 0.05) and increased levels of desulfovibrios (P < 0.05).

Table 3.

Effect of anti-inflammatory drugs on total eubacteria over the time period^a

Therapy	Mean log10 bacterial cells/mg ± SEM at time point (mo)
	0	3	6	9	12
None (n = 6)	6.6 ± 0.20	6.7 ± 0.15	7.3 ± 0.21	6.8 ± 0.31	6.9 ± 0.11
Balsalazide (n = 5)	6.7 ± 0.02	7.1 ± 0.21	7.3 ± 0.20	7.4 ± 0.31	7.0 ± 0.25
Salazopyrine (n = 5)	6.8 ± 0.23	7.4 ± 0.15	7.3 ± 0.21	7.1 ± 0.27	7.0 ± 0.21
Mesalamine (n = 14)	6.6 ± 0.24	6.9 ± 0.21	6.9 ± 0.15	7.1 ± 0.21	7.1 ± 0.17
Olsalazine (n = 1)	5.8 ± 0.30	7.2 ± 0.31	7.3 ± 0.21	7.3 ± 0.23	7.6 ± 0.15
Mesalazine (n = 2)	ND	7.2 ± 0.24	7.4 ± 0.34	6.9 ± 0.12	7.8 ± 0.31
Aza-Bal (n = 2)	6.8 ± 0.31	7.7 ± 0.35	6.9 ± 0.25	7.3 ± 0.15	7.7 ± 0.32
Pred-Aza-Sal (n = 1)	ND	ND	7.2 ± 0.32	ND	7.0 ± 0.21

^aValues are mean log₁₀ bacterial cells/mg from biopsy tissue ± standard errors of the means obtained from at least three biopsy specimens for patients receiving the same medication for a particular period of follow-up. Three patients changed therapy from mesalamine to mesalazine, balsalazide (Bal), or salazopyrine (Sal) during the course of the study. ND, not determined. Aza, azathioprine. Pred, prednisolene.

Table 4.

Effects of therapy on bacterial composition in UC patients over the study period^a

Bacterium	Mean log10 bacterial cells/mg ± SEM with therapy
	None	Balsalazide	Salazopyrine	Mesalamine	Olsalazine	Mesalazine	Aza-Bal	Pred-Aza-Sal
All eubacteria	6.9 ± 0.15	7.2 ± 0.17	7.0 ± 0.12	7.0 ± 0.1	7.4 ± 0.32	7.6 ± 0.27	7.0 ± 0.22	7.1 ± 0.05
Bacteroides genus	5.4 ± 0.16	5.7 ± 0.16	5.7 ± 0.13	5.6 ± 0.06	5.2 ± 0.65	5.7 ± 0.5	5.9 ± 0.23	6.0 ± 0.1
Bifidobacteria	4.7 ± 0.3	5.3 ± 0.3	5.8 ± 0.32	5.0 ± 0.12	5.3 ± 0.73	4.6 ± 0.43	4.9 ± 0.31	4.9 ± 0.06
E. rectale	5.0 ± 0.21	5.3 ± 0.22	6.0 ± 0.2	5.0 ± 0.15	4.5 ± 0.35	5.6 ± 0.27	5.1 ± 0.23	4.6 ± 0.1
F. prausnitzii	4.7 ± 0.21	5.4 ± 0.25	5.8 ± 0.25	5.0 ± 0.12	4.4 ± 0.35	5.5 ± 0.42	5.7 ± 0.55	4.3 ± 0.15
C. perfringens 2	4.6 ± 0.08	4.7 ± 0.08	4.9 ± 0.04	4.8 ± 0.04	4.6 ± 0.12	4.7 ± 0.12	4.5 ± 0.1	4.8 ± 0.03
C. clostridioforme	4.5 ± 0.25	4.8 ± 0.21	5.1 ± 0.2	4.5 ± 0.1	3.9 ± 0.32	4.8 ± 0.3	4.4 ± 0.27	4.2 ± 0.11
Enterobacteria	4.1 ± 0.1	4.2 ± 0.12	4.0 ± 0.08	4.3 ± 0.07	4.0 ± 0.27	3.9 ± 0.1	5.0 ± 0.3	4.5 ± 0.22
B. vulgatus	5.1 ± 0.15	4.5 ± 0.12	4.7 ± 0.1	4.7 ± 0.07	4.6 ± 0.4	4.7 ± 0.3	5.6 ± 0.3	4.0 ± 0.12
Ent. faecalis	4.1 ± 0.1	3.8 ± 0.08	4.2 ± 0.08	4.0 ± 0.2	4.0 ± 0.1	3.9 ± 0.12	4.3 ± 0.08	4.0 ± 0.1
Lactobacilli	3.7 ± 0.08	4.1 ± 0.13	3.8 ± 0.15	3.8 ± 0.08	4.4 ± 0.05	4.2 ± 0.25	4.8 ± 0.3	3.7 ± 0.03
P. anaerobius	2.8 ± 0.1	2.8 ± 0.08	3.4 ± 0.12	2.8 ± 0.04	2.3 ± 0.16	3.2 ± 0.1	2.8 ± 0.2	3.6 ± 0.08
R. albus	3.2 ± 0.21	3.6 ± 0.3	2.9 ± 0.08	3.3 ± 0.1	2.9 ± 0.18	3.4 ± 0.2	3.1 ± 0.6	ND
Desulfovibrios	2.4 ± 0.2	3.0 ± 0.12	3.2 ± 0.12	3.1 ± 0.07	3.0 ± 0.05	2.6 ± 0.1	2.5 ± 0.17	3.6 ± 0.35
C. perfringens 1	2.6 ± 0.23	3.0 ± 0.2	3.5 ± 0.22	3.2 ± 0.1	3.5 ± 0.45	2.8 ± 0.1	3.3 ± 0.45	ND
C. butyricum	2.6 ± 0.12	2.5 ± 0.12	2.6 ± 0.15	2.7 ± 0.06	3.0 ± 0.3	2.7 ± 0.15	2.2 ± 0.21	2.7 ± 0.15

^aValues are mean log₁₀ bacterial cells/mg from biopsy tissue ± standard errors of the means obtained from at least three biopsy specimens for the patients on the same medication over the 1-year period of the study. Three patients changed therapy from mesalamine to mesalazine, balsalazide, or salazopyrine during the course of the study. The number of patients undergoing each therapy is shown in Table 3. ND, not detected.

Composition of UC mucosal biofilms in relation to disease severity.

CAI and SS reflect disease severity in UC. CAI provides information on bowel frequency, blood in stool, and general well-being, whereas SS gives the total sum of grades in friability, bleeding, vessel pattern, and overall appearance of the colon, indicating the extent of inflammation. Relationships between mucosal bacterial populations in UC patients during the study and measurements of CAI and SS were calculated by estimating the gradient of linear regression for bacterial density (log₁₀ cells/mg biopsy specimen) over the time period against each parameter. In Fig. 3A, the interaction of CAI and SS in relation to bacterial community structure is shown by a scattergram, defined by regressions of the best fit (first-order nonlinear regression). CAI increased with SS in 72% of cases. Data points on the regression lines represent coordinates of bacterial gradients, ranging from a highly negative (E. rectale) to a highly positive enterobacterial gradient. Mucosal densities of enterobacteria, desulfovibrios, bacteroides, and Ent. faecalis were correlated positively with CAI and SS, whereas there was a negative association with C. butyricum, R. albus, and E. rectale with these parameters. Lactobacilli and bifidobacteria were negatively related to CAI.

Fig 3.

Scattergram of mucosal bacterial gradients with respect to clinical activity index (CAI) in UC patients over the 12-month study period. Gradients were obtained from linear regression equations of the log-transformed number of bacterial cells per milligram of biopsy specimen against CAI and SS (A), age of UC patient (B), and duration of disease (C). Values for each data point in the regression curves are indicated as follows: 1, desulfovibrios; 2, bifidobacteria; 3, enterobacteria; 4, Eubacterium rectale; 5, Clostridium clostridioforme; 6, Faecalibacterium prausnitzii; 7, Clostridium butyricum; 8, Peptostreptococcus anaerobius; 9, Bacteroides genus; 10, lactobacilli; 11, Ruminococcus albus; 12, Clostridium perfringens 1; 13, Enterococcus faecalis; 14, Bacteroides vulgatus; 15, Clostridium perfringens 2; 16, total bacteria.

Relationship of mucosal bacteria with age and duration of disease.

The relationship of age and duration of disease with CAI or SS can be seen in Fig. 3B and C and 4. Both age and disease duration correlated negatively with CAI and SS; however, the results showed deviations in this relationship with different bacteria. Some mucosal organisms (e.g., E. rectale and C. clostridioforme) had a high age dependence, while the carriage of others, including enterobacteria and desulfovibrios, was less affected by this factor. The association with duration of disease was generally similar, in that E. rectale, lactobacilli, F. prausnitzii, R. albus, and C. perfringens correlated positively. Enterobacteria, bacteroides, desulfovibrios, Ent. faecalis, and C. perfringens had a positive relationship with both CAI and SS in the UC patients, while C. butyricum had a negative correlation. Bifidobacteria and lactobacilli exhibited a negative association with CAI but were positive with respect to SS.

Fig 4.

Scattergram of sigmoidoscopy scores (SS) and age (A) and duration of disease (B) gradients in relation to mucosal bacterial densities in UC patients during the course of the study. Gradients were obtained from linear regression equations of log-transformed numbers of bacterial cells per milligram of biopsy specimen against age and duration of disease, respectively (see the legend to Fig. 3).

DISCUSSION

To our knowledge, this study is the first to show variations in mucosal microbiota populations in patients with active disease at onset, over a 1-year time period, and relate them to changes in disease severity, age, and duration of disease. Molecular technologies involving analysis of the 16S subunit of the rRNA gene provide sensitive and reliable tools for studying intestinal bacteria (16, 18, 34); real-time PCR, which is a rapid, sensitive, high-throughput method, was used to quantitate and characterize the complex microbial communities on the gut mucosa. Specific PCR primers were chosen to cover an extensive range of intestinal bacteria at the group, genus, and species levels. To avoid disruption of the mucosal biofilm, no lavage treatments were used to prepare the bowel in the UC patients or controls prior to taking biopsy specimens.

Previous cultural and molecular studies indicated that a dysbiosis may occur in mucosal populations in UC (2, 3, 12, 35–39), and in this investigation, we found that distinct differences in the UC mucosal microbiota between patients with active UC and the non-IBD comparators occurred at baseline. Similar to a previous investigation (3), but in contrast to other studies, which found higher numbers of microorganisms in IBD patients than in controls (37, 38), bacteria were found in higher numbers on the mucosa of the non-IBD controls in this study. This may have been due to the method of processing of the biopsy samples, the fact that the subjects were not pretreated before colonoscopy, and differences in analytical methods employed. In IBD, bifidobacteria, lactobacilli, and butyrate-producing species (C. butyricum, E. rectale, and F. prausnitzii) showed the greatest reductions in cell numbers, with significantly increased levels of P. anaerobius occurring on UC mucosa compared to the control group. Comparison of bacterial communities in biopsy specimens from normal and inflamed colonic mucosae in patients with acute UC in a previous study also showed significant reductions in numbers of lactobacilli and bifidobacteria (35). Reduced numbers of lactobacilli in colonic biopsy specimens were also found in pouchitis patients, who were reported to have low numbers of lactobacilli and bifidobacteria in the gut lumen (40). Reduced numbers of epithelium-associated bifidobacteria were also found in patients with active and quiescent ulcerative colitis compared to normal controls (41), and in a recent investigation, low levels of mucosal bifidobacteria were detected in children with active UC (42). Low numbers of anti-inflammatory bifidobacteria and increased levels of highly immunogenic peptostreptococci (43) were found on the UC gut mucosa in a previous study using culturing techniques (3), leading to the hypothesis that these organisms might be associated with disease etiology. Indeed, a Bifidobacterium longum strain isolated from a healthy rectal mucosa, in combination with a prebiotic, was subsequently shown to strongly reduce mucosal inflammation in UC and Crohn's disease patients (44, 45).

The fecal and mucosal microbiotas of healthy individuals have been shown to be relatively stable over time (20, 46). Due to ethical constraints, it was not possible to obtain samples from the control patients over time; therefore, although we cannot be certain that no changes in mucosal populations occurred, their baseline levels were used for comparison with longitudinal variations in UC patients. We demonstrated that the microbiota varied in UC mucosal tissue over a 1-year period, with large variations being observed for some bacterial populations, particularly bifidobacteria, F. prausnitzii, and the E. rectale group. These observations involving changes in microbial populations led us to investigate how different organisms were related to CAI and SS and to determine whether their carriage was associated with the age of the patient or disease duration. Results indicated that differences between some mucosal species and disease severity were evident in UC, although age and disease duration were less significant factors. Mucosal cell population densities of enterobacteria, desulfovibrios, bacteroides, and Ent. faecalis correlated positively with CAI and SS.

Sulfate-reducing bacteria (SRB) were previously linked to IBD (47, 48), mainly because these organisms produce sulfide, which is toxic to colonic epithelial cells (49). This metabolite also inhibits butyrate metabolism in colonocytes (50) as well as phagocytosis and bacterial killing (51). SRB were detected on the colonic mucosa in UC patients and in individuals without any form of IBD in this study, and their numbers were reduced significantly on UC mucosa compared to controls only at baseline. Previous work detected no significant difference in SRB levels on the rectal mucosae of patients with UC and non-IBD controls and suggested that if these organisms are involved in UC, some host defect, possibly in sulfide detoxification pathways or in bacterial antigen handling, is needed for disease to occur (16). However, when the UC patients are compared in the present study, it can be seen that there is a link between numbers of SRB and disease severity in the UC patients (Fig. 3).

UC patients have increased antibody titers to bacteroides (52), particularly B. vulgatus (53, 54), and this organism has been linked to UC in animal studies (55). Similar to previous cultural work (3, 4) and fluorescence in situ hybridization (FISH) analyses (18), bacteroides were found to be the predominant colonizing species on the UC mucosa in this investigation at baseline. However, statistical analysis showed that B. vulgatus had no relationship with disease severity. In contrast, enterobacteria were an important bacterial group associated with disease severity in UC. These organisms were consistently linked with high CAI and SS (Fig. 3 and 4), although total numbers were higher in normal tissues than in UC tissue (Fig. 1). Further evidence of a role for these organisms in UC comes from randomized controlled trials in which treatment with a nonpathogenic Escherichia coli strain was compared with mesalazine treatment in patients either in remission over 12 weeks (56) or after relapse (57), over 12 months. Both studies found E. coli to be as effective as mesalazine in preventing relapse.

In conclusion, real-time PCR provided a sensitive and efficient method for investigating the profiles of a comprehensive range of mucosal bacteria in UC patients longitudinally over the study period. We carried out an extensive analysis of mucosal bacterial populations in ulcerative colitis patients over a 1-year period. Results showed that some organisms, particularly bifidobacteria, peptostreptococci, and desulfovibrios, as well as putative pathogens, such as C. perfringens, might be linked in different ways to colitis severity. Numbers of putative health-promoting organisms such as bifidobacteria, lactobacilli, and butyrate-producing species (C. butyricum, E. rectale, and F. prausnitzii) were reduced in active UC and showed the greatest increase over the period of investigation. Moreover, the study showed that high numbers of enterobacteria on the mucosa were linked with high severity grades in UC, particularly with SS. However, further work is needed to determine treatment effects and whether changes in mucosal colonization by these organisms initiate or are a consequence of gut inflammation.